On Wed, Apr 25 2018, Brian Callahan <[email protected]> wrote: > On 04/25/18 09:31, Jeremie Courreges-Anglas wrote: >> On Mon, Apr 23 2018, Brian Callahan <[email protected]> wrote: >>> Hi ports -- >>> >>> Attached is an update to MuseScore. Wondering if there's any Ninja gurus >>> here who can figure out the circular dependencies going on (see diff). >>> I had to use USE_NINJA=No in order to get MuseScore to build. >>> >>> In case there isn't a good Ninja solution, ok? >> Not a cmake/ninja guru but I thought I'd take a look at their pull >> requests, and I found this: >> >> https://github.com/musescore/MuseScore/pull/2772 >> >> and more precisely: >> >> Fix ninja build errors due to mops1 and mops2 dependency cycles >> >> dependency cycle: mops1 -> CMakeFiles/mops1 -> all.h -> mops1 >> dependency cycle: all.h -> mops2 -> CMakeFiles/mops2 -> all.h.gch -> all.h >> >> >> https://github.com/musescore/MuseScore/pull/2772/commits/810586f52e21b1ac1d15ae81af2a1f809f4ea4b2 >> >> I'm not sure that just commenting those deps is the solution, but at >> least you have some data. Maybe just renaming the dependencies would be >> enough? >> > > I must be losing my mind. I swear I tried that and it didn't work. But > of course it does work. > Here's the updated diff. > > OK?
ok jca@; I only built it once with ninja, if it breaks due to random build order you can still set USE_NINJA=No and take your time to find the proper cmake fix. -- jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
